U.S. patent application number 14/126576 was filed with the patent office on 2014-04-24 for sit-to-stand transfer detection.
This patent application is currently assigned to KONINKLIJKE PHILIPS N.V.. The applicant listed for this patent is KONINKLIJKE PHILIPS N.V.. Invention is credited to Heribert Baldus, Stephan Schlumbohm, Wei Zhang.
Application Number | 20140114604 14/126576 |
Document ID | / |
Family ID | 46508141 |
Filed Date | 2014-04-24 |
United States Patent
Application |
20140114604 |
Kind Code |
A1 |
Zhang; Wei ; et al. |
April 24, 2014 |
SIT-TO-STAND TRANSFER DETECTION
Abstract
There is provided a method for identifying a sit-to-stand
transfer in measurements of the movement of a user, the method
comprising obtaining measurements of the vertical acceleration
experienced by the user during movement; obtaining measurements
indicating changes in height of a part of the user during movement;
processing the measurements of the vertical acceleration to
identify candidate movements corresponding to a sit- to-stand
transfer by the user; and determining an identified candidate
movement as a sit-to-stand transfer where the identified candidate
movement coincides with an increase in height. A corresponding
apparatus and computer program product are also provided.
Inventors: |
Zhang; Wei; (EINDHOVEN,
NL) ; Baldus; Heribert; (Aachen, DE) ;
Schlumbohm; Stephan; (FRANKFURT, DE) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
KONINKLIJKE PHILIPS N.V. |
EINDHOVEN |
|
NL |
|
|
Assignee: |
KONINKLIJKE PHILIPS N.V.
EINDHOVEN
NL
|
Family ID: |
46508141 |
Appl. No.: |
14/126576 |
Filed: |
June 19, 2012 |
PCT Filed: |
June 19, 2012 |
PCT NO: |
PCT/IB2012/053083 |
371 Date: |
December 16, 2013 |
Current U.S.
Class: |
702/141 |
Current CPC
Class: |
A61B 5/6822 20130101;
A61B 5/1116 20130101; A61B 2562/0219 20130101; A61B 5/7282
20130101; G01P 15/00 20130101 |
Class at
Publication: |
702/141 |
International
Class: |
G01P 15/00 20060101
G01P015/00 |
Foreign Application Data
Date |
Code |
Application Number |
Jun 28, 2011 |
EP |
11171700.5 |
Claims
1. A method for identifying a sit-to-stand transfer in measurements
of the movement of a user, the method comprising: obtaining
measurements of the vertical acceleration experienced by the user
during movement; obtaining measurements indicating changes in
height of a part of the user during movement; processing the
measurements of the vertical acceleration to identify candidate
movements corresponding to a sit-to-stand transfer by the user by
matching the measurements of the vertical acceleration to a
predetermined acceleration profile for a sit-to-stand transfer; and
identifying a candidate movement corresponding to a sit-to-stand
transfer in the result of the step of matching where there is a
peak, a first local minimum within a predetermined time period
before the identified peak and a second local minimum within a
predetermined time period after the identified peak; and
determining an identified candidate movement as a sit-to-stand
transfer where the identified candidate movement coincides with an
increase in height.
2. (canceled)
3. (canceled)
4. A method as claimed in claim 3, wherein the candidate movement
corresponding to a sit-to-stand transfer by the user is identified
in the result of the step of matching where the peak has a
magnitude in a predetermined range.
5. A method as claimed in claim 4, wherein a candidate movement
corresponding to a sit-to-stand transfer is further identified
where: (i) the difference between the magnitude of the peak and the
magnitude of the first local minimum is less than a first threshold
value; (ii) the difference between the magnitude of the peak and
the magnitude of the second local minimum is less than a second
threshold value; and (iii) the magnitude of the second local
minimum after the peak is less than the magnitude of the first
local minimum.
6. A method as claimed in claim 3, wherein the step of determining
an identified candidate movement as a sit-to-stand transfer
comprises: identifying a first sample in the measurements
indicating changes in height that corresponds to a first sample,
s1, in the result of the step of matching that is before the first
local minimum that exceeds a first threshold value; identifying a
second sample in the measurements indicating changes in height that
corresponds to a first sample, s2, in the result of the step of
matching that is after the second local minimum that exceeds a
second threshold value; and determining the change in height of the
part of the user from the identified first and second samples.
7. A method as claimed in claim 6, wherein the step of determining
the change in height from the first and second samples comprises:
determining the average of the height of the part of the user over
an evaluation window ending with the first sample; determining the
average of the height of the part of the user over an evaluation
window beginning with the second sample; and subtracting the two
averages to give the change in height during the candidate
sit-to-stand transfer.
8. A method as claimed in claim 3, the method further comprising
the steps of: estimating the variation of the vertical
acceleration; and determining the timing of the start and/or the
end of the identified sit-to-stand transfer in the measurements of
the vertical acceleration using the estimated variation.
9. A method as claimed in claim 8, wherein the step of determining
the timing of the start of the identified sit-to-stand transfer
comprises: identifying a sample in the estimated variation that
occurs before the first local minimum in the result of the step of
matching and that that is below a third threshold value, the sample
indicating the start of the identified sit-to-stand transfer.
10. A method as claimed in claim 8, wherein the step of determining
the timing of the end of the identified sit-to-stand transfer
comprises: identifying a sample, s1, in the result of the step of
matching that is before the first local minimum that exceeds a
first threshold value; identifying a sample, s2, in the result of
the step of matching that is after the second local minimum that
exceeds a second threshold value; identifying the lowest value in
the measurements of the vertical acceleration between s1 and s2;
and identifying the first sample after the lowest value in the
measurements of the vertical acceleration that exceeds a fifth
threshold value, said sample indicating the end of the identified
sit-to-stand transfer.
11. A method as claimed in claim 1, wherein the step of obtaining
measurements of the vertical acceleration experienced by the user
during movement comprises: obtaining three-dimensional measurements
of the acceleration experienced by the user during movement; and
processing the three-dimensional measurements to estimate the
vertical acceleration experienced by the user.
12. A method of determining the power used during a sit-to-stand
transfer by a user, the method comprising: identifying a
sit-to-stand transfer in measurements of the movement of a user
according to the method as claimed in claim 1; processing the
measurements of the vertical acceleration to determine an estimate
of the power used during the sit-to-stand transfer.
13. A method of determining a risk of falling for a user, the
method comprising: determining the power used during a sit-to-stand
transfer by a user as claimed in claim 12; determining a risk of
falling for the user from the determined power.
14. A computer program product, comprising computer program code
that, when executed on a computer or processor, causes the computer
or processor to identify a sit-to-stand transfer in measurements of
the movement of a user by: obtaining measurements of the vertical
acceleration experienced by the user during movement; obtaining
measurements indicating changes in height of a part of the user
during movement; processing the measurements of the vertical
acceleration to identify candidate movements corresponding to a
sit-to-stand transfer by the user by: matching the measurements of
the vertical acceleration to a predetermined acceleration profile
for a sit-to-stand transfer; and identifying a candidate movement
corresponding to a sit-to-stand transfer in the result of the step
of matching where there is a peak, a first local minimum within a
predetermined time period before the identified peak and a second
local minimum within a predetermined time period after the
identified peak; and determining an identified candidate movement
as a sit-to-stand transfer where the identified candidate movement
coincides with an increase in height.
15. An apparatus for identifying a sit-to-stand transfer in
measurements of the movement of a user, the apparatus comprising: a
processor for processing measurements of vertical acceleration
experienced by a user during movement to identify candidate
movements corresponding to a sit-to-stand transfer by the user, and
to determine an identified candidate movement as a sit-to-stand
transfer where the identified candidate movement coincides with a
measured increase in height, wherein the processor is configured to
identify candidate movements corresponding to a sit-to-stand
transfer by the user by matching the measurements of the vertical
acceleration to a predetermined acceleration profile for a
sit-to-stand transfer, and identifying a candidate movement
corresponding to a sit-to-stand transfer in the result of the step
of matching where there is a peak, a first local minimum within a
predetermined time period before the identified peak and a second
local minimum within a predetermined time period after the
identified peak.
16. An apparatus for identifying a sit-to-stand transfer in
measurements of the movement of a user, the apparatus comprising: a
processor for processing measurements of vertical acceleration
experienced by a user during movement to identify candidate
movements corresponding to a sit-to-stand transfer by the user, and
to determine an identified candidate movement as a sit-to-stand
transfer where the identified candidate movement coincides with a
measured increase in height, wherein the processor is configured to
identify candidate movements corresponding to a sit-to-stand
transfer by the user by matching the measurements of the vertical
acceleration to a predetermined acceleration profile for a
sit-to-stand transfer, and identifying a candidate movement
corresponding to a sit-to-stand transfer in the result of the step
of matching where there is a peak, a first local minimum within a
predetermined time period before the identified peak and a second
local minimum within a predetermined time period after the
identified peak.
Description
TECHNICAL FIELD OF THE INVENTION
[0001] The invention relates to a method and apparatus for
identifying a sit-to-stand transfer movement by a user.
BACKGROUND TO THE INVENTION
[0002] Falls are one of the greatest health risk factors for
elderly people. About one third of older people above the age of 65
fall at least once a year.
[0003] Many of these falls could be avoided by early identification
of fall risk and the application of effective and targeted fall
prevention programs.
[0004] Fall prevention trials based on strength and balance
training (SBT) have shown that the risk of falling for elderly
people can be reduced. Balance performance measures can be used as
early indicators of fall risk, and also to measure the progress of
fall prevention programs. The `sit-to-stand` (STS) transfer has
been identified as one important movement which can be used as a
balance performance measure. Domain experts can compare the graph
of the power generated during a sit-to-stand transfer for fall
prevention with the ECG graph in cardiovascular disorders. In daily
life, a person performs the STS transfer many times a day.
[0005] Conventionally, only clinical measurement systems (such as
those including a force plate and an optical marker system) allow
an accurate quantification of power during a sit-to-stand transfer.
In these measurement systems, the force plate provides the vertical
ground reaction force and the optical marker system provides a
measure of displacement in three dimensions. The combination of
both measurements is used to quantify the power during a
sit-to-stand transfer.
[0006] These measurement systems have several drawbacks. Firstly,
they are clinical equipment, which requires the user to attend a
clinic. Preparing for and performing measurements is labor
intensive (particularly if optical markers need to be attached to
specific parts of the body). In addition, they only provide a
snapshot of the user's balance performance, where, owing to the
clinical setting, the user commonly performs above their average
capability. Finally, the measurement systems involve a procedure
which is quite cumbersome for the user.
[0007] WO 2010/035187 entitled "Power Measurement and Apparatus"
discusses an apparatus for estimating the peak power used by a user
in performing the vertical component of a movement, such as a
sit-to-stand transfer, the apparatus comprising an accelerometer
for attachment to a user and for measuring the acceleration
experienced by a user; the apparatus further comprising a processor
configured to receive the measurements of the acceleration from the
accelerometer attached to the user; estimate the vertical
accelerations from the received measurements; and estimate the
power used from the vertical accelerations.
[0008] Existing activity monitoring technologies identify postures
or movements by classifying a sequence of sensor data of tens of
seconds or minutes in length. However, it is difficult to
accurately detect a sit-to-stand transfer that is typically
completed within 2 or 3 seconds.
[0009] Therefore, there is a need for a method and apparatus that
can identify such a transfer from measurements of the movement of a
user, so that the power used by the user in performing the movement
can be calculated. There is also a need for a method and apparatus
that can detect the onset and end of the transfer within a certain
degree of accuracy in order for the power analysis to provide
useful results.
SUMMARY OF THE INVENTION
[0010] According to a first aspect of the invention, there is
provided a method for identifying a sit-to-stand transfer in
measurements of the movement of a user, the method comprising
obtaining measurements of the vertical acceleration experienced by
the user during movement; obtaining measurements indicating changes
in height of a part of the user during movement; processing the
measurements of the vertical acceleration to identify candidate
movements corresponding to a sit-to-stand transfer by the user; and
determining an identified candidate movement as a sit-to-stand
transfer where the identified candidate movement coincides with an
increase in height.
[0011] According to a preferred embodiment, the step of processing
the measurements of the vertical acceleration to identify candidate
movements corresponding to a sit-to-stand transfer by the user
comprises matching the measurements of the vertical acceleration to
a predetermined acceleration profile for a sit-to-stand
transfer.
[0012] Preferably, a candidate movement corresponding to a
sit-to-stand transfer by the user is identified in the result of
the step of matching where there is a peak, a first local minimum
within a predetermined time period before the identified peak and a
second local minimum within a predetermined time period after the
identified peak.
[0013] Furthermore, the candidate movement corresponding to a
sit-to-stand transfer by the user is preferably identified in the
result of the step of matching where the peak has a magnitude in a
predetermined range.
[0014] In a yet further preferred embodiment, a candidate movement
corresponding to a sit-to-stand transfer is further identified
where (i) the difference between the magnitude of the peak and the
magnitude of the first local minimum is less than a first threshold
value; (ii) the difference between the magnitude of the peak and
the magnitude of the second local minimum is less than a second
threshold value; and (iii) the magnitude of the second local
minimum after the peak is less than the magnitude of the first
local minimum.
[0015] In one embodiment, the step of determining an identified
candidate movement as a sit-to-stand transfer comprises identifying
a first sample in the measurements indicating changes in height
that corresponds to a first sample, s1, in the result of the step
of matching that is before the first local minimum that exceeds a
first threshold value; identifying a second sample in the
measurements indicating changes in height that corresponds to a
first sample, s2, in the result of the step of matching that is
after the second local minimum that exceeds a second threshold
value; and determining the change in height of the part of the user
from the identified first and second samples.
[0016] In that embodiment, the step of determining the change in
height from the first and second samples comprises determining the
average of the height of the part of the user over an evaluation
window ending with the first sample; determining the average of the
height of the part of the user over an evaluation window beginning
with the second sample; and subtracting the two averages to give
the change in height during the candidate sit-to-stand
transfer.
[0017] In some embodiments, a more precise estimate of the start
and end of the sit to stand transfer can be found by estimating the
variation of the vertical acceleration; and determining the timing
of the start and/or the end of the identified sit-to-stand transfer
in the measurements of the vertical acceleration using the
estimated variation.
[0018] Preferably, the step of determining the timing of the start
of the identified sit-to-stand transfer comprises identifying a
sample in the estimated variation that occurs before the first
local minimum in the result of the step of matching and that that
is below a third threshold value, the sample indicating the start
of the identified sit-to-stand transfer.
[0019] Preferably, the step of determining the timing of the end of
the identified sit-to-stand transfer comprises identifying a
sample, s1, in the result of the step of matching that is before
the first local minimum that exceeds a first threshold value;
identifying a sample, s2, in the result of the step of matching
that is after the second local minimum that exceeds a second
threshold value; identifying the lowest value in the measurements
of the vertical acceleration between s1 and s2; and identifying the
first sample after the lowest value in the measurements of the
vertical acceleration that exceeds a fifth threshold value, said
sample indicating the end of the identified sit-to-stand
transfer.
[0020] In a preferred embodiment, the step of obtaining
measurements of the vertical acceleration experienced by the user
during movement comprises obtaining three-dimensional measurements
of the acceleration experienced by the user during movement; and
processing the three-dimensional measurements to estimate the
vertical acceleration experienced by the user.
[0021] According to a second aspect of the invention, there is
provided a method of determining the power used during a
sit-to-stand transfer by a user, the method comprising identifying
a sit-to-stand transfer in measurements of the movement of a user
according to the method described above; and processing the
measurements of the vertical acceleration to determine an estimate
of the power used during the sit-to-stand transfer.
[0022] According to a third aspect of the invention, there is
provided a method of determining a risk of falling for a user, the
method comprising determining the power used during a sit-to-stand
transfer by a user as described above; and determining a risk of
falling for the user from the determined power.
[0023] According to a fourth aspect of the invention, there is
provided a computer program product, comprising computer program
code that, when executed on a computer or processor, causes the
computer or processor to identify a sit-to-stand transfer in
measurements of the movement of a user as described above. Further
computer program products are provided that cause a computer or
processor to execute a method of determining the power used during
a sit-to-stand transfer by a user and a method of determining a
risk of falling for a user as described above.
[0024] According to a fifth aspect of the invention, there is
provided an apparatus for identifying a sit-to-stand transfer in
measurements of the movement of a user, the apparatus comprising a
processor for processing measurements of vertical acceleration
experienced by a user during movement to identify candidate
movements corresponding to a sit-to-stand transfer by the user, and
to determine an identified candidate movement as a sit-to-stand
transfer where the identified candidate movement coincides with a
measured increase in height.
[0025] According to a preferred embodiment, the processor is
configured to identify candidate movements corresponding to a
sit-to-stand transfer by the user by matching the measurements of
the vertical acceleration to a predetermined acceleration profile
for a sit-to-stand transfer.
[0026] Preferably, the processor is configured to identify a
candidate movement corresponding to a sit-to-stand transfer by the
user in the result of the matching where there is a peak, a first
local minimum within a predetermined time period before the
identified peak and a second local minimum within a predetermined
time period after the identified peak.
[0027] Preferably, the processor is configured to identify a
candidate movement corresponding to a sit-to-stand transfer by the
user in the result of the matching where the peak has a magnitude
in a predetermined range.
[0028] In a yet further preferred embodiment, the processor is
further configured to identify a candidate movement corresponding
to a sit-to-stand transfer where (i) the difference between the
magnitude of the peak and the magnitude of the first local minimum
is less than a first threshold value; (ii) the difference between
the magnitude of the peak and the magnitude of the second local
minimum is less than a second threshold value; and (iii) the
magnitude of the second local minimum after the peak is less than
the magnitude of the first local minimum.
[0029] In one embodiment, the processor is configured to determine
an identified candidate movement as a sit-to-stand transfer by
identifying a first sample in the measurements indicating changes
in height that corresponds to a first sample, s1, in the result of
the step of matching that is before the first local minimum that
exceeds a first threshold value; identifying a second sample in the
measurements indicating changes in height that corresponds to a
first sample, s2, in the result of the step of matching that is
after the second local minimum that exceeds a second threshold
value; and determining the change in height of the part of the user
from the identified first and second samples.
[0030] In that embodiment, the processor is configured to determine
the change in height from the first and second samples by
determining the average of the height of the part of the user over
an evaluation window ending with the first sample; determining the
average of the height of the part of the user over an evaluation
window beginning with the second sample; and subtracting the two
averages to give the change in height during the candidate
sit-to-stand transfer.
[0031] In some embodiments, a more precise estimate of the start
and end of the sit to stand transfer can be found where the
processor is configured to estimate the variation of the vertical
acceleration; and determine the timing of the start and/or the end
of the identified sit-to-stand transfer in the measurements of the
vertical acceleration using the estimated variation.
[0032] Preferably, the processor is configured to determine the
timing of the start of the identified sit-to-stand transfer by
identifying a sample in the estimated variation that occurs before
the first local minimum in the result of the matching and that that
is below a third threshold value, the sample indicating the start
of the identified sit-to-stand transfer.
[0033] Preferably, the processor is configured to determine the
timing of the end of the identified sit-to-stand transfer by
identifying a sample, s1, in the result of the matching that is
before the first local minimum that exceeds a first threshold
value; identifying a sample, s2, in the result of the matching that
is after the second local minimum that exceeds a second threshold
value; identifying the lowest value in the measurements of the
vertical acceleration between s1 and s2; and identifying the first
sample after the lowest value in the measurements of the vertical
acceleration that exceeds a fifth threshold value, said sample
indicating the end of the identified sit-to-stand transfer.
[0034] In a preferred embodiment, the processor is configured to
obtain three-dimensional measurements of the acceleration
experienced by the user during movement; and to process the
three-dimensional measurements to estimate the vertical
acceleration experienced by the user.
[0035] According to a further embodiment, the apparatus is for
determining the power used during a sit-to-stand transfer by a
user, wherein the processor in the apparatus is further configured
to identify a sit-to-stand transfer in measurements of the movement
of a user; and to process the measurements of the vertical
acceleration to determine an estimate of the power used during the
sit-to-stand transfer.
[0036] According to a yet further embodiment, the apparatus is for
determining a risk of falling for a user, wherein the processor in
the apparatus is further configured to determine the power used
during a sit-to-stand transfer by a user; and to determine a risk
of falling for the user from the determined power.
BRIEF DESCRIPTION OF THE DRAWINGS
[0037] Embodiments of the invention will now be described, by way
of example only, with reference to the following drawings, in
which:
[0038] FIG. 1 shows a sensor unit in accordance with an embodiment
of the invention attached to a user;
[0039] FIG. 2 is a block diagram of a sensor unit in accordance
with an embodiment of the invention;
[0040] FIG. 3 is a flowchart illustrating a method for identifying
a sit-to-stand transfer in measurements of the movement of a
user;
[0041] FIG. 4 is a graph illustrating an example of the variation
in vertical acceleration during a sit-to-stand transfer;
[0042] FIG. 5 is a block diagram illustrating an algorithm for
detecting a sit-to-stand transfer;
[0043] FIG. 6 shows the input signals to the algorithm and the
signals obtained during some of the processing steps; and
[0044] FIG. 7 illustrates an exemplary matched filter which has
been optimized for use in detecting a sit-to-stand transfer.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0045] As shown in FIG. 1, the invention provides a sensor unit 2
that is to be worn by the user 4. In the illustrated embodiment,
the sensor unit 2 is provided in the form of a pendant with a neck
cord 6 for placement around the user's neck. Alternatively, the
sensor unit 2 can be configured to be worn at or on a different
part of the user's body, such as the trunk, pelvis or sternum, and
will comprise a suitable arrangement for attaching the sensor unit
2 to that part of the body (for example a belt or a strap if the
unit 2 is attached to the pelvis or sternum).
[0046] The sensor unit 2 is used to measure the movement of the
user 4 and to process the measurements to determine when the user 4
has executed a sit-to-stand transfer. In preferred embodiments, the
sensor unit 2 is also used to determine the power or strength used
during the sit-to-stand transfer from the measurements of the
movement of the body of the user 4. Alternatively, this processing
can be performed in a base unit that is separate to the sensor unit
2 worn by the user 4 (not shown in FIG. 1).
[0047] FIG. 2 shows a preferred embodiment of the sensor unit 2 in
accordance with the invention. The sensor unit 2 comprises an
accelerometer 8 that measures acceleration along three orthogonal
axes and a sensor 9 that measures the altitude or height of the
sensor unit 2 above the ground (or more particularly that measures
changes in the altitude or height of the sensor unit 2 above the
ground, or enables those changes to be measured). The sensor 9 for
measuring the altitude or height of the sensor unit 2 can comprise,
for example, an altimeter or air pressure sensor, although those
skilled in the art will be aware of other types of sensors that can
be used. The signals output by the accelerometer 8 and sensor 9 are
provided to a processor 10 for analysis.
[0048] The sensor unit 2 also comprises a memory 12 and a
transmitter or transceiver circuitry 14. The memory 12 is used for
storing measurements from the accelerometer 8 and sensor 9, and for
storing the results of the analysis by the processor 10. The
transmitter or transceiver circuitry 14 is used for transmitting
the results of the analysis to a remote (base) unit or a computer
where they can be viewed or studied by the user or a healthcare
provider.
[0049] In some embodiments, the accelerometer 8 is a
micro-electromechanical system (MEMS) accelerometer. The
acceleration experienced by the accelerometer 8 can be sampled at a
rate of 50 Hz, although it will be appreciated that many other
sampling frequencies can be used. Where sensor 9 is an air pressure
sensor or altimeter, the measurements of the height of the sensor
unit 2 above the ground can be sampled at a frequency of around 1.8
Hz, although again it will be appreciated that other sampling
frequencies can be used.
[0050] Depending on the particular type of sensor used for the
sensor 9 for measuring height, the sensor 9 may output signals
indicative of the height above the ground (or sea level in the case
of an air pressure sensor), in which case the time series of height
measurements can be analyzed by the processor 10 to determine the
change in height from one measurement sample to the next (or over a
predetermined number of measurement samples). Alternatively, the
sensor 9 can directly output an indication of the change in height
of the sensor unit 2 from the previous or an earlier specified
measurement sample.
[0051] In an embodiment of the invention, the measurements
collected by the accelerometer 8 and sensor 9 are analyzed by the
processor 10 in the sensor device 2 to determine the occurrence of
a sit-to-stand transfer, and optionally the power or peak power
exerted by the user in performing the transfer. Alternatively, the
measurements from the accelerometer 8 and sensor 9 could be
transmitted to a base unit via the transmitter/transceiver
circuitry 14, with the base unit analyzing the measurements to
determine the occurrence of a sit-to-stand transfer. In either
case, the processing can be performed in (near) real-time or the
measurements from the accelerometer 8 and the sensor 9 can be
stored in the memory 12 or the base unit for future processing
(i.e. offline).
[0052] FIG. 3 shows a flowchart illustrating the steps required to
identify a sit-to-stand transfer in measurements of the movement of
the user. Firstly (step 101), measurements of the acceleration
experienced by the sensor unit 2 (and therefore the user 4, since
the user is wearing the sensor unit 2) are obtained. Secondly (step
103), measurements of changes in the height of the sensor unit 2
(and therefore the part of the user 4 that the sensor unit 2 is
attached to) above the ground are obtained. The measurements of the
acceleration and height (or changes in height) are obtained over
substantially the same period of time.
[0053] Next, in step 105, the measurements of the acceleration are
processed to identify movements in the measurements that may
correspond to a sit-to-stand transfer by the user 4. The parts of
the accelerometer measurement (i.e. a sequence of measurement
samples) that are identified in this step as possibly corresponding
to a sit-to-stand transfer are termed `candidate movements`.
[0054] In a preferred embodiment of the invention, the candidate
movements are identified by matching the measurements of the
acceleration to an acceleration profile that is expected to occur
during a sit-to-stand transfer.
[0055] The graph in FIG. 4 shows the acceleration measured in the
vertical direction during a typical sit-to-stand motion. The user 4
starts from rest (i.e. the measured acceleration in the vertical
direction is approximately 0) and the user begins to move at time
t.sub.s. The acceleration measured at this time is denoted
Acc.sub.vert.sub.--.sub.s. There is typically a small minimum in
the acceleration profile just after the user starts to move and
before they rise off their chair. Subsequently, the user's hip
leaves the means of support (i.e. chair) at time t.sub.ho (`ho`
represents hip off), and the acceleration at this time is denoted
Acc.sub.vert.sub.--.sub.ho. The acceleration in the vertical
direction then increases to a peak (the peak reaction) denoted
Acc.sub.vert.sub.--.sub.pr at time t.sub.pr. The peak reaction is
followed by the lowest reaction which is a negative acceleration
denoted Acc.sub.vert.sub.--.sub.lr occurring at time t.sub.lr. The
end of the movement occurs at time t.sub.e, with the acceleration
denoted Acc.sub.vert.sub.--.sub.e.
[0056] Thus, in step 105 of the flowchart in FIG. 3, the candidate
movements are identified by analyzing the accelerometer
measurements to identify sequences of samples whose profile match
or substantially match the profile shown in FIG. 4.
[0057] In step 107, the change in height occurring during each
candidate movement is determined from the measurements obtained in
step 103, and sit-to-stand transfers are determined to have
occurred where any identified candidate movement coincides with a
change in height that is within a predetermined range. The
predetermined range encompasses the height changes expected to
occur during a typical sit-to-stand transfer, which for example can
correspond generally to length of the user's thigh. In this case,
the lower bound for the range can be around 0.1 or 0.2 meters, for
example, and the upper bound for the range can be set to a value of
0.6, 0.75, 0.8 or 1 meter, for example. It will be appreciated that
the threshold can be personalized to the height or thigh length of
the user and can also be set taking into account the resolution of
the height or altitude measurements provided by the sensor 9.
[0058] It will also be appreciated that sit-to-stand transfers can
alternatively be determined by comparing the change in height to a
threshold value, with a sit-to-stand transfer being identified
where the change in height exceeds the threshold value. In this
case, the threshold can correspond to the lower bound for the
predetermined range described above. However, this embodiment may
result in a higher false positive identification rate than the
range embodiment described above, since activities such as climbing
the stairs may be identified as a sit-to-stand transfer (whereas
this movement would be discarded as a possible sit-to-stand
transfer by the upper bound of 0.6-1 meter in the range
embodiment).
[0059] A more detailed algorithm for detecting a sit-to-stand
transfer in accordance with the invention and for determining the
timing of the transfer is shown in FIG. 5. The algorithm takes as
an input the three-dimensional acceleration signal measured by the
accelerometer 8 (which comprises a separate signal for each of the
three axes of the accelerometer 8) and an air pressure measurement
from air pressure sensor 9.
[0060] The initial part of the algorithm, represented by blocks 19,
20, 21, 22 and 23, is a pre-processing stage in which the
accelerometer and pressure sensor signals are processed for use in
the subsequent analysis stages of the algorithm. Firstly, the 3D
acceleration signals from the accelerometer 8 are low-pass filtered
(block 19) to remove noise which could affect the accuracy of the
subsequent processing. In one embodiment, a Butterworth low-pass
filter with a cut-off frequency of 2 Hz is applied to the signals
from each of the three axes of the accelerometer 8. Alternatively,
it would be possible to apply different filter characteristics such
as a Chebyshev low-pass filter or other types of filter known to
those skilled in the art. It will also be appreciated that the
cut-off frequency of 2 Hz could be varied dependent on the
particular characteristics of the noise from the accelerometer
8.
[0061] As the orientation of the sensor unit 2 relative to the
fixed reference frame (such as the Earth) in which the user 4 moves
can change (particularly where the sensor unit 2 is in the form of
a pendant), it is necessary to process the measurements from the
accelerometer 8 to determine the vertical component of acceleration
experienced by the sensor unit 2 (and therefore user 4) during the
movement.
[0062] Therefore, the low-pass filtered 3D acceleration signals are
input to block 20 that estimates the vertical acceleration. The
vertical acceleration is denoted vert_acc.
[0063] One technique for estimating the vertical component of
acceleration from a 3D accelerometer signal having an arbitrary
orientation is described in WO 2010/035191, the content of which is
hereby incorporated by reference. Briefly, according to that
technique, the vertical component of acceleration is estimated from
measurements of acceleration acting on an accelerometer, the
accelerometer having an arbitrary orientation relative to the fixed
reference frame, by (i) examining the signals from the
accelerometer to identify the axis of the accelerometer having the
highest component of acceleration, (ii) determining the orientation
of the accelerometer by determining the angle between the
acceleration acting on the accelerometer (this acceleration being
assumed to be generally due to gravity) and the axis with the
highest component of acceleration and (iii) using the estimated
orientation of the accelerometer to determine the acceleration in
the vertical direction from the measurements of acceleration.
[0064] Those skilled in the art will be aware of other techniques
for estimating the vertical component of acceleration from the
measurements from a 3D accelerometer. For example, the sensor unit
2 can include a gyroscope for providing a signal indicating the
orientation of the sensor unit 2, and this signal can be used to
derive the vertical component of acceleration.
[0065] FIG. 6(a) shows an exemplary signal representing the
vertical acceleration obtained from measurements by a sensor unit 2
of a user performing a sit-to-stand transfer, walking for 3 meters
and then sitting back down, which was repeated three times. It can
be seen in FIG. 6(a) that there are three separate areas of
activity represented in the signal.
[0066] Another stage of the pre-processing concerns the calculation
of an estimate of the variation of acceleration. Firstly, a high
pass filter 21 is applied to each of the low-pass filtered 3D
acceleration signals in order to remove the DC component. In one
embodiment, a Butterworth high-pass filter with a cut-off frequency
of 0.5 Hz is used to remove the D.C. component in the acceleration
signals. It will be appreciated that another filter, for example a
Chebyshev high-pass filter or other types of filter known to those
skilled in the art could be used. It will also be appreciated that
a different cut-off frequency to 0.5 Hz could be chosen.
[0067] After high-pass filtering, the variation of the acceleration
is estimated in block 22. In a preferred embodiment, the standard
deviation of each of the three components of the 3D acceleration
signal is computed for a time t over a window of predetermined
length (for example, one second, although it will be appreciated
that another appropriately sized window could be used) and the
maximum standard deviation out of the three axes is identified. The
maximum standard deviation at time t is denoted max_std_acc and is
given by equation 1 below.
max_std_acc=max[std(acc.sub.--i(t-0.5,t+0.5)),i=x,y,z] (1)
[0068] FIG. 6(d) shows the standard deviation calculated for each
of the three axes of acceleration. In FIG. 6(d), line 40
corresponds to the x-axis accelerometer signal, line 42 corresponds
to the y-axis accelerometer signal, and line 44 corresponds to the
z-axis accelerometer signal.
[0069] A third pre-processing stage 23 estimates the altitude of
the sensor unit 2 from the measurements from the air pressure
sensor 9. As indicated above, the input to this stage 23 is the raw
air pressure signal p.sub.t from the air pressure sensor 3. As
mentioned previously, the air pressure can be sampled at a rate of
1.8 Hz (or in any case at a much lower sampling rate than the
acceleration signals). Therefore, the air pressure signal p.sub.t
is firstly upsampled to match the sampling rate (e.g. 50 Hz) of the
acceleration signals (the upsampled pressure signal is denoted
p.sub.t'). The altitude at time t (denoted alt_t) can then be
estimated from the air pressure sensor measurements using equation
2 below:
alt.sub.--t=44330*(1-p.sub.t'/101325).sup.0.19 (2)
[0070] Equation (2) is derived from the air pressure to altitude
conversion function shown in equation (3):
alt_t = T 0 L ( 1 - ( p p 0 ) RL gM ) ( 3 ) ##EQU00001##
Where:
TABLE-US-00001 [0071] Symbol Quantity Typical Value alt_t Altitude
in meters p Air pressure p.sub.0 Standard atmospheric pressure
101325 kPa at sea level L Temperature lapse rate 0.0065 Km.sup.-1
T.sub.0 Standard temperature at sea level 288.15 K g Gravitational
acceleration at 9.80665 ms.sup.-2 Earth's surface M Molar mass of
dry air 0.0289644 kg mol.sup.-1 R Universal gas constant 8.31447 J
mol.sup.-1 K.sup.-1
[0072] The resulting altitude signal is then smoothed, preferably
with a median filter having a predetermined length, for example of
around 3 seconds. The filter is applied to the time series of
estimated altitudes, resulting in a smoothed altitude signal
alt_meas which is output from the altitude estimation stage 23, as
shown in FIG. 6(c). In FIG. 6(c), the y-axis represents altitude in
meters relative to sea level.
[0073] It will be appreciated that in alternative embodiments of
the invention where a different type of altitude, height or change
in height sensor is used, processing stage 23 may be adapted or
omitted as appropriate.
[0074] Following the pre-processing of the input signals, various
features are extracted in order to determine if a sit-to-stand
transfer has occurred, and if so, the power of the user in
performing the sit-to-stand transfer.
[0075] Two main stages of feature extraction are required in order
to determine if a sit-to-stand transfer has occurred. The first
stage 24 of the feature extraction executes step 105 of the
flowchart in FIG. 3 and identifies the candidate movements in the
vert_acc signal. In particular, block 24 matches the vert_acc
signal to a predetermined pattern representing the vertical
acceleration that is expected to occur during a sit-to-stand
transfer.
[0076] In a preferred embodiment, the first stage 24 of the feature
extraction applies a matched filter having an impulse response that
approximates the vertical acceleration experienced during a
sit-to-stand transfer to the vertical acceleration signal
(vert_acc) output from the vertical acceleration estimation block
20. The output of the matched filter is a set of coefficients that
indicate the match of the measurements to the pattern. Each
coefficient represents the match of a number of consecutive
measurement samples (covering a time period of the same length as
the predetermined pattern) to the predetermined pattern. The higher
the coefficient, the better the match of the measurements to the
pattern (and therefore the greater the chance that a sit-to-stand
transfer has occurred). The filtered signal is denoted
vert_acc_matfilt and is shown in FIG. 6(b).
[0077] In a preferred embodiment, the matched filter used in block
24 can be as shown in FIG. 7, which has been optimized to detect a
sit-to-stand transfer. The matched filter shown in FIG. 7 excludes
gravity (9.8 ms.sup.-2) The first curve 50 shows a typical vertical
acceleration pattern of a sit-to-stand transfer. The second curve
51 shows an applied matched filter characteristic that approximates
the first curve 50. It will be appreciated that the matched filter
characteristic may be expressed using many different functions, but
in this embodiment, the matched filter characteristic is given by
equation 4 below.
A.sub.1sin c[W.sub.1(t-t.sub.1)]+A.sub.2sin c[W.sub.2(t-t.sub.2)]
(4)
[0078] This characteristic is a combination of two sin c functions
with scale parameters defined in p. p is a parameter vector with
six elements:
[A.sub.1,A.sub.2,W.sub.1,W.sub.2,t.sub.1,t.sub.2] (5)
[0079] Each entry in p defines a different scale parameter. A.sub.1
and A.sub.2 are amplitude scale parameters, which define the peak
deviation of the two sin c waves respectively. The parameters
W.sub.1 and W.sub.2 are frequency scale parameters, which define
the frequency of the two sin c waves. The parameters t.sub.1 and
t.sub.2 are phase scale parameters, which define the position of
the sin c waves. The values of the six elements in the parameter
vector p are set to tune the function of the matched filter to the
sit-to-stand transfer characteristic 50 in FIG. 7.
[0080] It will be appreciated that the values of the elements of
the parameter vector p can be provided by many known curve-fitting
methods. In one case, the desired parameters could be calculated by
applying a nonlinear least-squares regression algorithm, however
many other types of fitting algorithms are well known in the art
and could be applied. The nonlinear least-squares regression
algorithm generates different parameter combinations corresponding
to different functions. The generated functions are then fitted to
the data set of desired patterns according to a least-squared error
criterion. When the function yields a minimum value of least square
error among the combination of parameters, an optimized fit has
been found.
[0081] After matched filtering, the filtered signal is processed to
identify movements that may correspond to a sit-to-stand transfer
by the user. The processing consists of firstly identifying any
peak having a magnitude in a predetermined range in the
vert_acc_matfilt signal. In the exemplary signal shown in FIG.
6(d), peaks whose magnitudes are in the range of 110 to 200 are
identified. It will be appreciated that this part of the processing
can alternatively comprise identifying any peak having a magnitude
above a threshold value in the vert_acc_matfilt signal. In this
case, the threshold can correspond to the lower bound for the
predetermined range described above. However, this classification
may result in a higher false positive identification rate than the
range embodiment described above.
[0082] For each identified peak, the algorithm attempts to identify
respective local minima occurring within a predetermined time
period before and after the identified peak in the vert_acc_matfilt
signal. In the exemplary signal shown in FIG. 6(b), the algorithm
looks for local minima within a period of 2 seconds before and
after the identified peak. If no local minima are identified for a
particular peak, that peak of the vert_acc_matfilt signal is not
considered to correspond to a sit-to-stand transfer.
[0083] Finally, a candidate movement corresponding to a
sit-to-stand transfer is identified as a peak having the required
local minima and at which the difference between the magnitude of
the peak and the magnitude of the local minimum before the peak is
less than a first threshold value, the difference between the
magnitude of the peak and the local minimum after the peak is less
than a second threshold value, and the magnitude of the local
minimum after the peak is less than the magnitude of the local
minimum before the peak.
[0084] In simplified implementations of the invention, the
magnitude requirements applied to the local minima can be relaxed,
with the algorithm simply identifying the peak, the magnitude of
the peak, and the presence of local minima before and after the
peak.
[0085] In the exemplary signal shown in FIG. 6(b), the first
threshold is 25 and the second threshold is 200. It will be
appreciated that the values chosen for the first and second
thresholds are tuned to an experimental dataset, and different
threshold values could be used.
[0086] It can be seen in FIG. 6(b) that four possible movements
have been highlighted as candidate sit-to-stand transfers,
occurring roughly at times 1.65, 1.69, 1.78 and 1.87.
[0087] As described above with reference to step 107 of FIG. 3,
candidate sit-to-stand transfers are identified as actual
sit-to-stand transfers when they occur at the same time as a change
in the height of the sensor unit 2 that is within a predetermined
range. Thus, block 25 determines the change in height or altitude
that has occurred during each candidate sit-to-stand transfer. In
order for block 25 to evaluate the altitude change of a candidate
sit-to-stand transfer identified in the matched filtering block 24,
block 25 receives a copy of the vert_acc_matfilt signal and
indications of which parts of the signal correspond to candidate
sit-to-stand transfers from the matched filtering block 24. Block
25 also receives the estimated altitude measurement signal,
alt_meas, from estimation block 23.
[0088] A candidate sit-to-stand transfer found in the output from
the matched filter 24 consists of three key samples. These are the
peak, the local minimum before the peak (min.sub.--1), and the
local minimum after the peak (min.sub.--2). These samples are
marked for one of the candidate sit-to-stand transfers in FIG.
6(b). In order to estimate the altitude change over the correct
time period, it is necessary to identify the right samples in the
altitude measurement signal.
[0089] Firstly, the nearest sample (s1) before the local minimum
before the peak (min.sub.--1) whose value is larger than a
threshold is found. Secondly, the nearest sample (s2) after the
local minimum after the peak (min.sub.--2) whose value is larger
than a threshold is found. It will be appreciated that
theoretically, this threshold should be g.sup.2; however in
practice, different values might be provided by the training
dataset due to slight inaccuracies in the accelerometer, for
example. In one embodiment, this threshold is 98.
[0090] The altitude change of the candidate sit-to-stand transfer
is then estimated as the difference between the altitudes at
samples s1 and s2.
[0091] Preferably, since there may be small fluctuations in the
altitude measurement (due to noise), the altitude change of the
candidate sit-to-stand transfer is estimated as the difference
between the mean of the altitude measurement over a time window
starting at the second local minimum, and the mean of the altitude
measurement over a time window ending at the first local minimum.
These time windows can be one second, although it will be
appreciated that windows of other lengths can be used. In equation
form, this can be expressed as
alt_diff=mean(alt_meas(s2:s2+t.sub.w))-mean(alt_meas(s1-t.sub.w:s1))
(6)
where t.sub.w is the length of the window. In this way, the mean
value of the altitude data one second before the start and one
second after the candidate transfer is evaluated. When a
sit-to-stand transfer has occurred, a lower altitude should be
observed before the transfer (when the user 4 is in the sitting
position) than the altitude observed after the transfer (when the
user 4 is in the standing position).
[0092] The output of the candidate sit-to-stand transfer
identification block 24 and the altitude change block 25 are
provided to a decision block 26 which determines whether any of the
candidates are sit-to-stand transfers. In particular, any candidate
movement occurring at the same time a change in altitude or height
within a predetermined range is deemed to be a sit-to-stand
transfer. The change in height should be an increase in height (by
definition of a sit-to-stand transfer), and the predetermined range
can be, for example, between 0.1 and 0.75 meters. As described
above with reference to step 107 of FIG. 3, the upper bound can be
omitted at the expense of a greater false positive detection
rate.
[0093] It can be seen in FIG. 6 that of the four candidate
movements highlighted in FIG. 6(b), the last three occur at the
same time as an increase in height that is in the range 0.1 to
0.75. Thus, the candidate movements at times 1.69, 1.78 and 1.87
are deemed to correspond to sit-to-stand transfers. The candidate
movement at time 1.65 coincides with a reduction in the measured
height and is therefore discarded. The algorithm then repeats for a
new set of input data (represented by block 27 in FIG. 5).
[0094] As described earlier, for detected sit-to-stand transfers
(block 28), the power used by the user 4 during the transfer can be
estimated. This is performed in block 29. In order for the estimate
to be as accurate as possible, it is necessary to determine the
timing of the start and end of the sit-to-stand transfer.
[0095] Therefore, a block 30 determines the timing of the
sit-to-stand transfer and receives inputs from the block 22 which
estimates of the variation of the acceleration and the vertical
acceleration profile after matched filtering, vert_acc_matfilt.
[0096] In a simple embodiment, s1 and s2 are used to identify the
start and end of the sit-to-stand transfer for the purposes of
calculating the power used.
[0097] However, as will be known to those skilled in the art, the
matched filter introduces a delay which is related to the number of
filter taps. This delay causes the candidate sit-to-stand transfer
to be delayed with respect to the actual onset of the sit-to-stand
transfer in the vert_acc_matfilt signal. Therefore, in a preferred
embodiment, the output of block 22 that estimates the variation in
acceleration, max_std_acc can be used to determine the actual onset
of a sit-to-stand transfer.
[0098] Firstly, the most adjacent sample in the signal max_std_acc
before s1 whose value is smaller than a threshold is identified.
This threshold determines where the onset of the actual
sit-to-stand transfer (denoted t_start) is found. In an exemplary
case the threshold may be 0.35, but it will be understood that
different threshold values smaller than 1 may be used, with the
specific value being selected, in part, based on the size of the
computing window being applied to the signal. Then, the largest
local minimum of the estimate of the vertical acceleration
(vert_acc) between s1 and s2 (in other words, the lowest value of
vert_acc between s1 and s2) is found. The most adjacent sample
after the largest local minimum of the estimate of the vertical
acceleration, whose value is larger than a threshold value, which
in a preferred embodiment is based on gravity (i.e. 9.8 ms.sup.-2),
is defined as the end of the actual sit-to-stand transfer (t_end).
The solid black bars in FIG. 6(b) and corresponding circles in FIG.
6(a) indicate t_start and t_end for each actual sit-to-stand
transfer. The values for t_start and t_end for each detected
sit-to-stand transfer are provided to power calculation block
29.
[0099] Block 29 also receives the vert_acc signal from block 20 and
calculates the peak power present in the sit-to-stand transfer. In
particular, the section of the estimate of the vertical
acceleration between the start and end of the sit-to-stand transfer
(i.e. between t_start, t_end) is isolated.
[0100] As described in WO 2010/035187, the peak power during a
sit-to-stand transfer can be calculated using:
Power(t)=m*(vert_acc(t)+g)*.intg..sub.t.sub.--.sub.start.sup.t.sup.--.su-
p.end(vert_acc*(t))dt (7)
where m is the mass of the user 4 and g is acceleration due to
gravity.
[0101] Following the computation of the power in the sit-to-stand
transfer by block 29, the result is output for further processing
or analysis.
[0102] It will be appreciated that the peak power output from the
power computation stage 29 could be stored enabling the evaluation
of the variation in power over a sustained period of time, such as
one month. The evaluation could be based on sit-to-stand peak power
in combination with other known parameters, such as the user's age,
gender, and health conditions. The evaluation could also be
performed in combination with parameters from other fall-related
assessments, such as time-up-and-go. If evaluation results pass a
fall-risk threshold, a caregiver or user could be alerted.
Alternatively or in addition, a report could be provided for
feedback on progress. Health professionals can obtain the same
report for the use of providing intervention services.
[0103] There is therefore provided a method and apparatus that can
identify a sit-to-stand transfer from measurements of the movement
of a user. This identification subsequently allows the power used
by the user in performing the movement can be calculated. In
addition, in certain embodiments, the method and apparatus detect
the onset and end of the transfer within a certain degree of
accuracy in order for the power analysis to provide useful
results.
[0104] While the invention has been illustrated and described in
detail in the drawings and foregoing description, such illustration
and description are to be considered illustrative or exemplary and
not restrictive; the invention is not limited to the disclosed
embodiments.
[0105] Variations to the disclosed embodiments can be understood
and effected by those skilled in the art in practicing the claimed
invention, from a study of the drawings, the disclosure and the
appended claims. In the claims, the word "comprising" does not
exclude other elements or steps, and the indefinite article "a" or
"an" does not exclude a plurality. A single processor or other unit
may fulfill the functions of several items recited in the claims.
The mere fact that certain measures are recited in mutually
different dependent claims does not indicate that a combination of
these measures cannot be used to advantage. A computer program may
be stored/distributed on a suitable medium, such as an optical
storage medium or a solid-state medium supplied together with or as
part of other hardware, but may also be distributed in other forms,
such as via the Internet or other wired or wireless
telecommunication systems. Any reference signs in the claims should
not be construed as limiting the scope.
* * * * *